From 3a4773c3c2bd0bbef244eb519b07208da9108e49 Mon Sep 17 00:00:00 2001
From: dcc <dcc@logografos.com>
Date: Sat, 2 Sep 2023 00:52:52 -0700
Subject: First

---
 .../web/rich_media/parsers/meta_tags_parser.ex     | 46 ++++++++++++++++++++++
 lib/pleroma/web/rich_media/parsers/o_embed.ex      | 29 ++++++++++++++
 lib/pleroma/web/rich_media/parsers/ogp.ex          | 10 +++++
 lib/pleroma/web/rich_media/parsers/twitter_card.ex | 15 +++++++
 4 files changed, 100 insertions(+)
 create mode 100644 lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex
 create mode 100644 lib/pleroma/web/rich_media/parsers/o_embed.ex
 create mode 100644 lib/pleroma/web/rich_media/parsers/ogp.ex
 create mode 100644 lib/pleroma/web/rich_media/parsers/twitter_card.ex

(limited to 'lib/pleroma/web/rich_media/parsers')

diff --git a/lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex b/lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex
new file mode 100644
index 0000000..320a5f5
--- /dev/null
+++ b/lib/pleroma/web/rich_media/parsers/meta_tags_parser.ex
@@ -0,0 +1,46 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.Web.RichMedia.Parsers.MetaTagsParser do
+  def parse(data, html, prefix, key_name, value_name \\ "content") do
+    html
+    |> get_elements(key_name, prefix)
+    |> Enum.reduce(data, fn el, acc ->
+      attributes = normalize_attributes(el, prefix, key_name, value_name)
+
+      Map.merge(acc, attributes)
+    end)
+    |> maybe_put_title(html)
+  end
+
+  defp get_elements(html, key_name, prefix) do
+    html |> Floki.find("meta[#{key_name}^='#{prefix}:']")
+  end
+
+  defp normalize_attributes(html_node, prefix, key_name, value_name) do
+    {_tag, attributes, _children} = html_node
+
+    data =
+      Map.new(attributes, fn {name, value} ->
+        {name, String.trim_leading(value, "#{prefix}:")}
+      end)
+
+    %{data[key_name] => data[value_name]}
+  end
+
+  defp maybe_put_title(%{"title" => _} = meta, _), do: meta
+
+  defp maybe_put_title(meta, html) when meta != %{} do
+    case get_page_title(html) do
+      "" -> meta
+      title -> Map.put_new(meta, "title", title)
+    end
+  end
+
+  defp maybe_put_title(meta, _), do: meta
+
+  defp get_page_title(html) do
+    Floki.find(html, "html head title") |> List.first() |> Floki.text()
+  end
+end
diff --git a/lib/pleroma/web/rich_media/parsers/o_embed.ex b/lib/pleroma/web/rich_media/parsers/o_embed.ex
new file mode 100644
index 0000000..0f30317
--- /dev/null
+++ b/lib/pleroma/web/rich_media/parsers/o_embed.ex
@@ -0,0 +1,29 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.Web.RichMedia.Parsers.OEmbed do
+  def parse(html, _data) do
+    with elements = [_ | _] <- get_discovery_data(html),
+         oembed_url when is_binary(oembed_url) <- get_oembed_url(elements),
+         {:ok, oembed_data = %{"html" => html}} <- get_oembed_data(oembed_url) do
+      %{oembed_data | "html" => Pleroma.HTML.filter_tags(html)}
+    else
+      _e -> %{}
+    end
+  end
+
+  defp get_discovery_data(html) do
+    html |> Floki.find("link[type='application/json+oembed']")
+  end
+
+  defp get_oembed_url([{"link", attributes, _children} | _]) do
+    Enum.find_value(attributes, fn {k, v} -> if k == "href", do: v end)
+  end
+
+  defp get_oembed_data(url) do
+    with {:ok, %Tesla.Env{body: json}} <- Pleroma.Web.RichMedia.Helpers.rich_media_get(url) do
+      Jason.decode(json)
+    end
+  end
+end
diff --git a/lib/pleroma/web/rich_media/parsers/ogp.ex b/lib/pleroma/web/rich_media/parsers/ogp.ex
new file mode 100644
index 0000000..b7f2b42
--- /dev/null
+++ b/lib/pleroma/web/rich_media/parsers/ogp.ex
@@ -0,0 +1,10 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.Web.RichMedia.Parsers.OGP do
+  @deprecated "OGP parser is deprecated. Use TwitterCard instead."
+  def parse(_html, _data) do
+    %{}
+  end
+end
diff --git a/lib/pleroma/web/rich_media/parsers/twitter_card.ex b/lib/pleroma/web/rich_media/parsers/twitter_card.ex
new file mode 100644
index 0000000..cc65372
--- /dev/null
+++ b/lib/pleroma/web/rich_media/parsers/twitter_card.ex
@@ -0,0 +1,15 @@
+# Pleroma: A lightweight social networking server
+# Copyright © 2017-2022 Pleroma Authors <https://pleroma.social/>
+# SPDX-License-Identifier: AGPL-3.0-only
+
+defmodule Pleroma.Web.RichMedia.Parsers.TwitterCard do
+  alias Pleroma.Web.RichMedia.Parsers.MetaTagsParser
+
+  @spec parse(list(), map()) :: map()
+  def parse(html, data) do
+    data
+    |> MetaTagsParser.parse(html, "og", "property")
+    |> MetaTagsParser.parse(html, "twitter", "name")
+    |> MetaTagsParser.parse(html, "twitter", "property")
+  end
+end
-- 
cgit v1.2.3